On Thu, May 17, 2018 at 09:34:47AM +0300, Alex Rosenbaum wrote: > When we discussed this in the past we agreed that while IBTA defines > the wire bits and must be compliant by all vendors, the SW > representation is Linux verbs specific and does not have to match > exactly to the IBTA wire definitions. No.. That is wrong - old kernels pass bits unmodified from PortInfo through this interface so any device that had IsOtherLocalChangesNoticeSupported set now also sets IP_BASED_GIDS This *might* have been OK because no roce devices probably set this bit, but it was still a bad choice. But now we are talking about adding an IB only bit that overlaps with IsLinkSpeedWidthPairsTableSupported, so any IB port on old kernels that was setting that bit will now be seen as setting GRH_REQUIRED, which is completely wrong uABI design. So, no. This new bit needs to go down a clean channel. There are two reserved bytes ib_uverbs_query_port_resp, I recommend converting one of them into flags instead. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
